Producing a mass storage backup using a log of write commands and time information
View Patent ↗Computer systems may lose data when a failure occurs within a system. To counteract such loss of data a backup system may be employed. Common backup systems make a copy of either of the data on a storage device or the data, which has changed, on a storage device. The process of backing up data may involve storing a relatively large amount of data and so is commonly done infrequently, such as once per day. If a computer's data is backed up only once per day, several hours of data may be lost if a computer system fails. Embodiments of the present invention may be used to prevent this type of data loss by backing up more frequently. In order to back up more frequently less data at a time is backed up. Instead of the data undergoing a wholesale backup infrequently, embodiments of the present invention form a timed log of the storage writes performed by the computer system. The log provides a running picture of activity to the computer storage system. By preserving the log, for example storing it at a remote site through a network connection, the state of the computer can be recreated with any desired granularity, by using the log entries to recreate the state of the data within the computer system at any desired time.
1. A method for backing up a mass storage the method comprising:
accepting, at a log-assisted disk, mass storage write commands for the mass storage to be backed up;
appending, at the log-assisted disk, a time to each of said mass storage write commands to form a log entry;
queuing log entries in a log queue at the log-assisted disk; and
writing log entries from the log queue into a log file in a local mass storage different from the mass storage to be backed up.
2. A method as in claim 1 further comprising storing the log file in a non volatile storage.
3. A method as in claim 2 wherein the local mass storage is a hard disk.
4. A method as in claim 2 wherein the storing the log file in a non volatile storage further comprises:
providing the log file to a network interface;
using the network interface to couple the log file into a network;
accepting the log file from the network; and
storing the log file on a mass storage device.
5. A method as in 4 wherein using the network interface to couple the log file into a network further comprises:
receiving a status from the network;
testing the status to determine if the network traffic is low; and
coupling the log file into the network dependant on the network traffic.
6. A method as in claim 1 the method further comprising taking a snapshot of the mass storage to be backed up prior to accepting mass storage write commands for the mass storage to be backed up.
7. A method as in claim 1 wherein the step of writing log entries from the log queue into a log file further comprises:
determining the sector to be written to from the most recent log entry;
searching for log entries having an earlier time stamp which writes to the same address; and
deleting any log entries with an earlier time stamp which writes data to the same address as the most recent log entry.
8. A method of recreating the state of a mass storage device at a given time the method comprising:
accepting a snapshot of the state of a mass storage device taken at a point in time earlier than the given time;
accepting log entries having timestamps later than the point in time of the snapshot;
writing the snapshot to a storage device on which the mass storage device is being restored;
writing the log entries having timestamps later than the point in time of the snapshot to the storage device on which the mass storage device is being restored; and
terminating the writing of the log entries when the timestamp of the log entry is equal to the given time.
9. A method as in claim 8 wherein the accepting a snapshot of the state of a mass storage device and accepting log entries from the time of the snapshot further comprises accepting a snapshot of the state of a mass storage device and accepting log entries from the point in time of the snapshot from a network connection.
10. A method as in claim 9 where the network is the Internet.
11. An article of manufacture comprising a computer readable media and computer code which causes a computer to:
accept mass storage write commands for a mass storage to be backed up at a log-assisted disk;
append a time to each of said mass storage write commands to form a log entry at the log-assisted disk;
queue log entries in a log queue at the log-assisted disk; and
write log entries from the log queue into a log file in a mass storage different from the mass storage to be backed up.
12. A method for continuously backing up computer data to a mass storage system, said method comprising:
a) receiving write commands from an application running on a computer;
b) converting each received write command to a sector write having a sector address and sector data;
c) sending each sector write to a log-assisted disk;
d) combining each sector write with a time stamp at the log-assisted disk to form a log entry;
e) queuing log entries at the log-assisted disk;
f) communicating the log entries to a mass storage system; and
g) storing the log entries in a log file at the mass storage system.
13. The method of claim 12 further including taking a snapshot of the data stored on the computer prior to receiving write commands.
14. The method of claim 12 wherein the mass storage system is in network communication with the computer.
15. The method of claim 14 wherein communicating the log entries to the mass storage system is further defined by:
a) providing the log entries to a network interface at the computer;
b) using the network interface to couple the log entries into a network; and
c) accepting the log entries from the network at the mass storage system.
16. The method of claim 12 wherein the mass storage system is located at the computer.
17. The method of claim 12 wherein the step of storing the log entries in a log file is further defined by:
a) determining a sector address to be written to from a received log entry;
b) searching for log entries having an earlier time stamp which were written to the same sector address; and
c) deleting any log entries with an earlier time stamp which were written to the same sector address as the received log entry.